Extension of fast R-CNN (Recurrent Convolution Neural Network) [1] and Recurrent-CNN [2] is Faster Recurrent CNN detection techniques. All these three strategies use convolutional neural networks (CNN). The distinction between them is the way to select the regions to work and the way those regions are categorised. Recurrent CNN and fast R-CNN used algorithm proposed region as a pre-processing step before running CNN. Algorithms are usually technical proposals such as Borders [3] or Selective Search [4] boxes, which are independent of CNN. Fast Recurrent CNN, the use of these systems becomes bottleneck process with relevance to CNN’s operation. Fast R-CNN solves this downside by applying the projected mechanism to use the region and then CNN. An RPN is a fully convex network that predicts object boundaries and object scores at each location simultaneously. © 2020, Springer Nature Singapore Pte Ltd.